In the previous post, we looked at Steve Yegge’s post where he made a projection from the vibe coding of today to controlling AI agent fleets in the near future that take over all coding reliably. Pondering this projection as a possible future, we realized that this future is not what we need as our actual problems in software development do not lie in a lack of developer productivity (or to be more precise: developer efficiency) but…

This blog series will be a bit different. It might leave you with more questions than answers and I apologize for that upfront. It is also more controversial than most other posts, I have written before. In this post, I will discuss the current developments regarding AI in software development more from a CTO’s perspective. This means, I cannot simply reason about the pros and cons of a topic but I also need to take the market forces into account.

In the previous post, we discussed how the broken feedback loops that software development projects create lead to a continuously deteriorating IT system landscape, resulting in an ever-shrinking dependability – which is probably the by far most important runtime property of software.

What is a project?

In the previous post, we discussed the biggest challenges, IT organizations face these days. We have seen four big challenges, each with its own peculiarities: